Case study

Pokerbook.

A mobile-first poker session tracker for casual home games.

Track buy-ins, settle the math and share the result with the group. A live product I designed and built solo.

Used by 50+ people · self-reported

Product, Design, BuildNext.jsSupabasePWALive Product
Pokerbook active session “Rana”: five players with buy-in counts and +1 buy-in controls, a ₹2,000 buy-in, 8 total buy-ins, a ₹16,000 table bank, a copy-link button and an add-player tile
The Problem

Poker nights are fun. Settlements are not.

Tracking, payouts, and group sharing usually happen across notes, calculators, and WhatsApp.

Tracking gets messy during the game

Re-buys happen mid-conversation, chips move fast, and paper notes drift out of date.

Settlement math becomes annoying

Hosts are left turning cash-outs into balances while everyone is ready to leave.

Results are hard to share cleanly

Final numbers usually end up as screenshots, partial notes, or a long chat explanation.

What it does

Built for the person running the table

One host sets up the night. Everyone else follows along from a link.

Setup

Share one link

User need
Everyone wants to see the numbers
Pokerbook
Copyable session link, so every player can follow the table from their phone.
Setup

Control who edits

User need
Only trusted hands on the ledger
Pokerbook
Intended experience: a host-set edit PIN for editing, view-only for everyone else.
Live

Track buy-ins and cash-outs

User need
Re-buys happen mid-hand
Pokerbook
Record changes as the game evolves, one tap at a time.
Settle

Settle clearly

User need
Know who owes what
Pokerbook
Settlement calculation turns final stacks into clear amounts owed.
Settle

Make mismatches explicit

User need
The totals don’t add up
Pokerbook
When buy-ins and cash-outs differ, the host corrects counts or reviews a proposed adjustment.
Share

Send it to the group

User need
End the night in the chat
Pokerbook
One-tap WhatsApp summary of the final settlement.
Product Decisions

What shaped the experience

Four calls about trust, money and where the group already talks.

Money

Settlement as the core value

The product earns trust when it turns noisy table activity into clear balances.

Trust

A PIN instead of accounts

The intended experience: one host-set PIN for editing, a read-only view for everyone else. Light trust at the table without a sign-up wall, not a security guarantee.

Edge case

Make mismatches explicit

Real tables rarely balance exactly. The host corrects counts or reviews a proposed adjustment. The reviewed version refuses invalid values and adjustments that would need negative counts.

Distribution

WhatsApp-native sharing

The final artifact is built for the chat where the group already coordinates.

Proposed metrics

What I'd measure

Proposed, not outcomes. No results are claimed here.

Proposed

Session creation completion rate

Why it matters

Shows whether hosts can start a game without friction.

Proposed

Settlement completion rate

Why it matters

Shows whether users reach the main payoff moment.

Proposed

Repeat host usage

Why it matters

Shows whether hosts come back for future games.

Tradeoffs

What I intentionally did not build

The MVP stays focused on solving one workflow well.

No login-first experience

Reduced setup friction for casual games, with identity deferred until there is repeat value.

No tournament mode in MVP

Cash games were the clearest wedge because settlement pain is immediate and frequent.

No social feed or community layer

Kept attention on the table workflow instead of adding another place to maintain presence.

Lesson

Money changes should be understandable and recoverable

The latest reliability pass was implemented with Claude Code and independently reviewed.

Say what changed

A money-related edit should show what it did before it is applied.

Refuse what cannot be right

Invalid values and impossible adjustments are blocked, not quietly patched.

Make it recoverable

When something goes wrong at the table, the host can see it and fix it.

A reviewed reliability update is on a separate branch; database rollout and deployment are pending confirmation.